import cv2
import mediapipe as mp
import gradio as gr
import numpy as np

# Initialize MediaPipe Hands
mp_hands = mp.solutions.hands
mp_drawing = mp.solutions.drawing_utils
hands = mp_hands.Hands(max_num_hands=1, min_detection_confidence=0.7)

def classify_gesture(landmarks):
    if not landmarks:
        return "Unknown"

    thumb_tip = landmarks[4]
    index_tip = landmarks[8]

    if thumb_tip.y < index_tip.y:
        return "A (Thumb Up)"
    return "Unknown"

def process_frame(frame):
    frame = cv2.flip(frame, 1)
    rgb = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
    result = hands.process(rgb)

    gesture = "No hand detected"

    if result.multi_hand_landmarks:
        for hand_landmarks in result.multi_hand_landmarks:
            mp_drawing.draw_landmarks(frame, hand_landmarks, mp_hands.HAND_CONNECTIONS)
            gesture = classify_gesture(hand_landmarks.landmark)

    cv2.putText(frame, f"Sign: {gesture}", (10, 40), cv2.FONT_HERSHEY_SIMPLEX, 1, (0, 255, 0), 2)
    return frame

# Gradio interface (compatible with old Gradio versions on Spaces)
demo = gr.Interface(
    fn=process_frame,
    inputs=gr.Image(type="numpy", tool="editor"),  # webcam snapshot button works here
    outputs=gr.Image(type="numpy"),
    title="Sign Language Recognition",
    description="Take a webcam snapshot or upload an image to detect hand signs"
)

demo.launch()